The development of biomechanical performance criteria so far has been directed at assessing specific thresholds for certain injuries, and converting these to numerical values of force and time measured on a dummy in a standard crash. In real life, however, occupants vary greatly in age, size, and tolerance to injury, and collisions vary from the slightest to the most severe. The use of a single size dummy in a single test clearly cannot fully reflect the range of variables present in the real world.

This paper examines some of the data already published on these variables. These are the distributions of severities of impacts, of types of collision, the age, sex, and size of occupants exposed to risk in the three designated seats in the car (driver, front passenger, rear passenger), the variation in injury severity with age and the relation of age to mortality for a given injury.

Some conclusions are drawn on the weightings which might be given to different severities of injuries, and some observations are made on the consequences of "under" and "over" protection. Suggestions for necessary further research projects are outlined briefly.